CaptainToBin well I am just happy if I can put backports in with brain switched off so keeping the original layout. For suite stuff I can then switch it on again.

Upps CaptainTobin without a capital B :)

tomman: how are you my good and dear vague acquaintance on the internet

and WG9s

is it me, or the Internet Archive navbar on archived sites is now broken on SeaMonkey?

(it now appears centered, lacking its styles, and I can't even get the Inspector to find any of its elements on the document tree!)

have a link to check?

sample link: web.archive.org/web/19961220113701/http://www.windows95.com/drivers

but really, load any page on the Wayback Machine

the toolbar the IA appends to every page should be floating on the top, there should even be a X button to close it

but since.. yesterday?, it now floats vertically centered, lacking all of its styles

it certainly depends on javascript, I think I don't get the toolbar, and the whole wayback machine has been changing to rely more and more on javascript :-\

here is the rub: I see no JS error this time :/

could be some CSS shenanigans instead?

heh, no universal VESA driver, or that's WDM-only?

the behavior of the Inspector is weird, too

now to see how it used to be... I'm already regretting asking this...

njsg: for that you would need to travel back to the present :)

does the wayback machine have some archived page from the wayback machine?

...good question

INCEPTION!

hm, I'm getting it at the top, what am I doing wrong

TypeError: t.getRootNode is not a function

i.imgur.com/aX009bJ.png here is how the toolbar renders on my end

...wait

franstam google webcomponents it seems.

franstam: thanks for the tip!

indeed it's WebComponents

have the two magical prefs enabled... and found the first site that breaks here

disabled both and now the Wayback Machine renders properly

(still no related error messages on dev console)

ah yes, with webcomponents I see the same as in your screenshot

guess the IA is one of the rare places having fallbacks in place for "not Chrome"?

...fittingly, considering the goals of the site

yeah, DOM Inspector doesn't seem to like the site either

Might need Shadow DOM.

I usually just don't have javascript enabled for the wayback machine

with dom.webcomponents.customelements.enabled and dom.webcomponents.enabled, the toolbar breaks (and so does the inspector)

with both disabled (as it is the default), everything works OK

...why do they want to use Google WebComponents® there!?

probably because they're in the "let's use javascript just because" idea

how do i tell whether a site is using google webcomponents or other naughty bits?

I think listing archived stuff under a path now requires javascript, for example

but even more amazingly, they keep compatibility for "non-mainstream" browsers

i dont really understand how to tell

franstam: take a look at the dev console

from those dev console messages

is there a list of messages that point to certain things?

if you find errors like "getRootNode is not a function" or "window.customElements is not defined", that's a dead giveaway

to be honest, I've found the wayback machine to be less reliable regarding saving (archiving) pages, sometimes it won't load, sometimes it loads an older archived version, and while it won't show the newer one if you try to request it, that one might or might not pop up at a later point *sigh*

franstam: those two usually point to required WebComponents crap

less reliable, compared to more in the past.

while another recent enemy are Unicode-aware regexes

(anything with "invalid identity escape in regular expression")

njsg: at least it's now useful for archiving executables and other binaries

10 years ago it SUCKED at archiving, say, old Microsoft patches or ZIP files

where was that cd-rom icon, btw? at least in 98 that wasn't the CD icon used in explorer (well, shell32.dll, I guess), was it? but it appeared somewhere. I thought it'd be newer, but perhaps it's older?

either shell32.dll, or cool.dll

I think 98 used the one on bottom right, the one left to the icon that's not the My Computer icon in 98 (but is in 98's shell32.dll, which might actually be what the screenshot shows?) codeproject.com/KB/shell/selecticon/selecticon.gif

er, bottom left.

status meeting starts in 30 minutes - wiki.mozilla.org/SeaMonkey/StatusMeetings/2023-02-05

here for the meeting. tryiung to get first in the appearances list, but then perhaps it is based on best looking or bessed dressed. ;-)

oh, need to install some pending Debian updates here

gonna reboot in a few minutes

That was my lame idea of making a joke about order of appearance rather than order of joining. Of course order of joining would give an unfair advantage to Trills! ;-)

OK I am done for making lame jokes to try to entertain during the wait for meeting start time.

brb, rebooting~

njsg or perhaps a photo on me and lots of photoshoping

or i need to get the latest google phone. the tv ad says it will make all of my photos look better!

status meeting starts in 15 minutes - wiki.mozilla.org/SeaMonkey/StatusMeetings/2023-02-05

status meeting starts in 5 minutes - wiki.mozilla.org/SeaMonkey/StatusMeetings/2023-02-05

# Appears as NJSG.

(to conclude the comic chat joke; here for the meeting, greetings .*!)

Hi rsx11m

hi frg (not Away as it appears ;-) )

Time for the status meeting - wiki.mozilla.org/SeaMonkey/StatusMeetings/2023-02-05

hi IanN also Away

hello world of JavaScript~

hi rsx11m

hi njsg WG9 tomman

Who's taking minutes?

me

thanks

Nominees for Friends of the Fish Tank?

tonymec|away for diligently testing 2.53 Linux 64-bit nightly builds.

I would like to nominate buc for digging out a clang > 13 fix and ewong for releasing the recent releases

seconding buc

seconding both

seconding all 3

hi tonymec

oh missed ewong was on same message I second or maybe third that

Hi tonymec

hi all

hi tonymec

wow, I didn't expect that. I just install WG9s's L64 builds as I see them published, and if something goes wrong, it won't get fixed unless reported, will it?

tonymec: yes, reporting helps get it fixed so you're being a good friend to the fish tank

seconding (fourthing?) all 3

tonymec we all do testing but some bugs are not even reported till the official release is out. Too few people test the perebeases and betas.

hi frg IanN WG9s rsx11m and anyone I might have forgotten

hi tonymec

i like my perebeases on toast

Action Items

tonymec: if no one actually tests them then how do we know if we boke something, or wy do we even bother to do the nightly builds good to have someone we know actaully runs them and lets us know if someting is borken.

I would like no typos but it is me so won't happen

WG9s: yeah, that's one of the reasons I test them

I don't think anything has changed on Action Items

no

Status of the SeaMonkey Infrastructure

bau

Nothing new here. Are we in agreement that we need a new builder for 2.53.17?

Linux builder

wait, what happened to .16?

tomman it will be released and then .17 is next

tomman: .16 b=with current builder and then update the linux version used to build for .17

oh, and what's changing for .17 on the builds?

toolchain stuff I guess?

i think just building using centos 7 is gatting harder to make work.

tomman we need later toolchains for macOS Linux and also python 3

Native macOS builds are broken becuase of patches which can't go in too.

got it

macOS 11 and up sdks also not usable and so on.

yeah, at least in the case of Linux, it looks like it's time to move on from CentOS 7

It will also mean a later libstdc requirement.

My linux builds use an antiquated 10 versions back fedora27 to keep our current backwards campatibility requirements

FWIW, current Debian stable (which is already ooooold) is based off GCC10's libstdc++6

perhaps also a later glabc requirement

glibc

no idea what glabc might be

Debian here has glibc 2.31

(2.24 on Stretch that it's on ELTS, 2.28 on Buster that it's on LTS, 2.31 on Bullseye)

what one targets CentOS 7?

3.4.19

but we currently target stdcxx 3.4.19 and glibc 2.17 which is same as what firefox targets

glibc is fine

Just stdcxx needs to move to 3.4.12 or .22 not sure. With the later toolchain and patches we might even become compatible with the older one again. Not sure if current Firefox still runs on EL7.

this is supposedly what current firefox requires:

STDCXX_MAX_VERSION = Version("3.4.19")

CXXABI_MAX_VERSION = Version("1.3.7")

GLIBC_MAX_VERSION = Version("2.17")

LIBGCC_MAX_VERSION = Version("4.8")

so what is listed as MAX is really the oldes version it will supposedly run on.

oh, so it's libstdc++6 here, and to make things more confusing, Debian packages them using GCC version numbers!

Status of the SeaMonkey Source Tree

tomman: those are the versions of the actual libraries and NOT the package versions.

IanN so agreed?

so what is in /usr/lib/libstdc++

Have some outstanding reviews and plan to do a few pushes to central soon.

But planned for Friday already so might be next weekend :)

bot then 6.0.24 maps to 3.4.24 etc

current centOS7 has 3.4.19

WG9s: ah, strings /path/to/libstdc++.so.6 reveals the truth

i have changed my 2.53 and central builds to use rust 1.67,0

GLIBCXX_3.4.28 here

WG9s 6.0.19 but tool low for the build tools.

i also changed central macOS cross compiles to sue MacOSX.sdk to version 13.1

1.67 works fine for 2.53 too. A clang fix made it into 2.53 so need to do a few test with 15 again.

As for 2.53.16 the current code will becom the base of the beta soon. So please test.

2.53.15 broke unintentionally on macOS 10.9 to 10.11 on macOS. Can probably not be fixed without backing out later stuff for Ventura so 10.12 will become the new minimum.

i have some builds posted at wg9s.com/macos and we asked for 10.9 testers but got no takers. the only build I am farly certain there will work is the one that backed out some angle version updates to sthat it would build using sdk 10.11 but that would be not the desired approach because there are issues building with that sdk and the latest mac OS

IF we could actually get someone to test one of the other approaches might be good to keep it working as long as possible

Would have liked to keep 10.11 supported a bit longer but any mac from the last 12 years can run at least 10.12 so not schedding much tears here.

Some older ones can even support in with hacks.

a nd the people who complained seem non-responsive now, but perhaps htat is becuase they gave up on us

Mac is getting even worse than Linux with regards of backwards compatibility

I am still building my latest 2.53 cross-compiles specifying -mmacosx-version-min=10.9 on the compile and link that does not seem to hurt anything and might make them rork for 10.9 to 10.11, but like I said can not find any testers.

I suppose we could change to doing our release builds specifying that but no idea how to tell if it is actually making any difference

Release Train

LInux got bad when the mozilla folks decided to no longer to static builds

2.53.15 is done. 2.53.16 beta will be next :)

As dicscussed previoulsly 2.57 from the esr60 source is basically dead. 2.53 suprpassed it already in features, functionality and stability.

^surpassed

Version number might be reused.

any plans to use one of the more recent ESR branches instead in the future?

or are they too far away?

only reason not an issue under windows is the whay it is packaged if you use the correct redist dif it includes the ddl files required in the package. fi the linux builds in clded the libc and libsddc++ required there would be no issue there either.

Extensions Tracking

My personal opionion is that the source did turly go to the dogs after 60. And even 60 was bad or we would reuse it. Sill keeping central updated

^truely

yepp :-/

the newer libc will work with older kernels because if try to use a feature not supported by the kernel there is a specific error that is returned and you can fall back to doing it the hard way

rsx11m build system is now on the way to 78 and other parts are between 60 and 91. Plus sec and other updates.

back when mozilla supported statically linking you could get the linux version to run under freebsd even though it was an entirely different kernel

WG9s well it would be a monster now so this option is gone.

well perhaps not had to do with 32 bit os and maximum file size issues

and other 32 bit os issues

kissue was the statically built binary got to be too big

that's because it's static ;-)

perhaps if it did not rub its feet on the carpet?

For extensions GreasMonkey got broken by 2.53.15. Needs at least locale updates internally.

Didn't test: forums.mozillazine.org/viewtopic.php?p=14948699#p14948699

back when only 32-bit maxiimum filesize was 2 gig and your app plus the os had to fit in 2 gig of memory

We unfortunately need the locale changes to stay to be able to do later updates.

Bug 1414390 is the culprit.

Palefill 1.23 still works with .15, that's the max we can go here

so we had dynamic linking and shared libraries stuff to get around this.

Flashblock seriously needs a rewrite (and a rename, since Flash is dead!), but can be made to work with .15 (although site whitelisting is now broken, need to figure out why)

tomman internal custom elements support is now pretty good bug regexp and shadow DOM is still urgkkk...

but not bug grumpf...

frg: yeah, webcomponents is no longer a big issue (other than with a few sites), but the Unicode regex stuff is the new enemy of the month

so many sites relying in that because frameworks do not care

I can only do my broken record speech and say that I am on it.

(and when you complain at framework devs, they deflect the blame to site devs)

it's sad that the packages that get picked up to be widely used are precisely those which don't seem to care about "what if we're used everywhere" and just cater to a few specific browsers...

there's "but this feature would help a lot with the kind of manipulation this framework was made to do" and then there's "we can't do basic error testing or add a fallback"

2.Next, Feature List, Planning & Roundtable

(does google translate's "translate this page" now detect when the newer performance features aren't present or does it still fails?)

s/fails/fail/

bau

AOB?

bau too

AOB: pass

okay, I'll call this a meeting, thanks for your time everyone

next meeting in 3 weeks, same bat channel, same bat time

see you then!

February 26?

exactly

yo. Expoect late meeting notes as usual from me.

hi stefanh! 8-)

just in time for the end of the meeting ...

rsx11m: Hi there!

Oh, right

mostly talking about build stuff and library versions

I know it's time to update to the next Debian release when I try to build stuff and fail with "time to update your compilers, buddy"

Hi stefanh

frg: Hi

hiya stefanh

stefanh we need to move macOS min requirements to 10.12 otherwise probably nothing which affects you personally :)

IanN: Hi there!

frg: heh, ok :-)

frg: I'm actually runjing 10.12.6

*running

stefanh some of the latest updates needed for Ventura compatibility play bad with 10.9 to 10.11

right

stefanh: hi

And native building is broken too but will be fxed for 2.53.17.

njsg: hi there

so wondering if like we do for windows we put the c runtime library in the application dir and that provides backward compability. i wouder if something similar can be done with glibc under linux

Something is wrong with SeaMonkey. Site by site is stopping work on it. Finally a week ago Facebook stopped too. What happens?